The combined effects of ENSO and Arctic Oscillation on wintertime fog days in eastern China

The data of fog days from weather observation stations in China and the NCEP/NCAR re-analysis data from 1954 to 2007 are used to investigate the combined effects of El Niño and Southern Oscillation (ENSO) and Arctic Oscillation (AO) on the number of winter fog days in eastern China. The results show that during El Niño, the enhanced low-level southwesterly warm-moist airflow could lead to the temperature rise and humidity increase in eastern China. Note that the rate of humidity increase is faster than the rate of temperature rise, which makes the air in eastern China easy to be saturated. Besides, in winter, North China is dominated by the sinking airflow, so a large-value area of fog days appears in eastern China with the center in North China. While in La Niña years, the atmospheric circulation and its influence on the fog days are the opposite. During the positive AO period, the East Asian trough weakens and the low-level westerly jet moves northward, preventing northwesterly cold air from moving southward. The warming and humidification of North China and the slight temperature drop in South China would cause more fog days in North China and fewer fog days in South China. The effect of negative AO is opposite to that of positive AO. The combined effects of ENSO and AO are far greater than the sum of their individual effects. Under El Niño and positive AO, the number of fog days would increase significantly in North China during the whole winter. Besides, ENSO has greater impacts than AO during early winter and vice versa during later winter.


Introduction
Fog is a common-seen weather phenomenon in the boundary layer. It refers to the water vapor condensations suspending in the atmosphere, such as water droplets and ice crystals. When the air temperature is close to the dew point, the water vapor in the air could condense into fog. According to the international definition, when the visibility is less than 1 km, it is defined as fog, and when the visibility is more than 1 km, it is defined as mist. Fog is a common disastrous weather in winter in eastern China. Fog damage is mainly reflected in the hazards to traffic safety and health. With the development of the national economy and the improvement of people's quality of life, the losses caused by fog are increasing day by day (Gultepe et al. 2007;Niu et al. 2010a;Cai et al. 2017).
There are many methods to study the climate characteristics of fog, and the most commonly used method is to count the local annual or monthly average fog days. The climatological characteristics of fog in different regions and the annual or seasonal fog days are mostly analyzed. The results show that fog in China has obvious seasonal variation and occurs most frequently in winter (Hu et al. 2020;Wang et al. 2005;Sun et al. 2013). In terms of spatial distribution, fog in China is mainly concentrated in the East and Southwest (Yu et al. 2019).
The formation of fog depends on the dynamic (transport and diffusion) and thermal conditions (temperature stratification conducive to water vapor condensation). Zhang et al. (2014) pointed out that the strong cold air activity in January 2013 is less, the East Asian trough is obviously weakened, and the East Asian monsoon is weak. This will lead to the abnormal high pressure of 500 hPa and inhibit the development of convection. On the other hand, there is a southerly wind anomaly (water vapor transport in the lower layer) at 925 hPa. The weakened wind speed and the obviously high temperature make it easy to form inversion which makes the atmosphere more stable. The stable atmosphere is conducive to the maintenance and development of fog episodes. In general, previous studies show that fog often occurs when the wind speed is low, the humidity is high, and the atmosphere is stable (Sachweh and Koepke 1997;Niu et al. 2010b).
Most of the researches are based on climate factors to discuss the conditions of fog, including the influence of climate factors on dynamic and thermodynamic conditions. In recent years, many scholars have begun to pay attention to the relationship between fog frequency and Arctic Oscillation (AO), Eurasian teleconnection, East Asian winter monsoon (EAWM), El Niño and Southern Oscillation (ENSO), and so on (Niu et al. 2010b;Yu et al. 2019;Li et al, 2019;Liu et al. 2020a). Li et al. (2016) found a close relationship between EAWM and fog days in China (Li et al. 2016). Under the background of global warming, the weakening of EAWM increases the frequency of fog days in China (Niu et al. 2010b). As the strongest tropical ocean signal, ENSO has a very important impact on the climate of different seasons in East Asia (Rasmusson and Wallace 1983;Philander 1990). Yu et al. (2019) found that ENSO exerts an important influence on the inter-annual variation of winter fogs in eastern China by affecting the southwesterly warmmoist airflow from the tropics. AO is the first mode of empirical orthogonal function (EOF) at 1000 hPa or sea-level pressure in the northern hemisphere, which is different from the highlatitude climate signal of the tropics. AO has a very significant impact on China's winter and spring (Chen and Zhou 2012;Chen et al. 2013;Cohen et al. 2014;Gong et al. 2001;He and Wang 2016;He et al. 2018;Lee and Jhun 2006;Li et al. 2014;Wang and Chen 2010;Wu and Wang 2002;Chen and Song 2019). Liu et al. (2020a) showed that AO affects the East Asian trough, which has a significant impact on the dynamic and thermal conditions of winter fogs in China.
Most of the current studies have focused on the effects of ENSO or AO on precipitation anomalies and other extreme weather events. But, the combined effects of ENSO and AO are likely to produce different results from those of the single factor (Chen et al. 2013). For example, in January 2008, South China suffered from severe snowstorms and cold and icy conditions, which was mainly caused by AO anomaly under the background of La Niña (Wen et al. 2009). When studied the causes of severe winter drought in Yunnan in 2009/2010, Song et al. (2011) found that AO was related to winter precipitation in Yunnan and was also modulated by ENSO. Zuo (2011) studied the relationship between AO and ENSO and its impact on the climate anomalies in China. In the different phase combinations of AO and ENSO, the precipitation anomaly in Southeast China is mainly controlled by ENSO. However, AO obviously adjusts the path of water vapor transport anomaly and the location of precipitation anomaly.
However, the combined effects of ENSO and AO on winter fog days in China are not clear. Therefore, in this paper, the inter-annual variability of winter fogs in eastern China is aimed. We attempt to explain how ENSO and AO jointly affect the number of fog days in eastern China and which of them is more important. The remainder of this paper is organized as follows. Section 2 describes the data and methodology. In Section 3, a special attention will be paid to the distinctive difference in fog frequency between early and late winter. Meanwhile, the climatological characteristics of fog   1958, 1958/1959, 1963/1964, 1965/1966, 1972/1973, 1977/1978, 1982/1983, 1986/1987, 1987/1988, 1991/1992, 1994/1995, 1997/1998, 2002/2003, 2006/2007 1954/1955,1955/1956, 1956/1957, 1964/1965, 1970/1971,1973/1974, 1975/1976, 1984/1985, 1988/1989, 1995/1996, 1998/1999, 1999/2000, 2000/2001, 2005/2006  days in eastern China are presented, and the major environmental factors controlling the fog formation are also discussed. In Section 4, we examine the composite patterns of fog days in early and late winter during El Niño, La Niña, +AO and −AO phases, and the associated large-scale dynamic and thermodynamic variables. Besides, physical mechanisms responsible for the distinctive patterns in early and late winter are also discussed in Section 4. In Section 5, we attempt to explain how ENSO and AO jointly affect the number of fog days in eastern China and which of them is more important. Finally, a summary and discussions are given in Section 6.

Data and methods
Previous studies have shown that the frequency of fog days in China is variable both in time and space. Fog events may occur frequently or rarely in a given year (Syed et al. 2012). The fog days were obtained from the surface observation dataset of 503 weather stations in China. From daily weather station records during the period of 1954-2007, we calculated fog days at each station for early winter (November and December, hereinafter referred to as ND) and late winter (January and February, hereinafter referred to as JF). The monthly data after interpolation with a spatial resolution of 1°× 1°was adopted. To highlight the characteristics of fog days in eastern China, the mapping range of fog days was set as (70-140°E, 15-55°N). The National Centers for Environmental Prediction/ National Center for Atmospheric Research (NCEP/NCAR) monthly re-analysis data was used in this paper, including wind, geopotential height, specific humidity, relative humidity, and temperature. The dew point temperature could be calculated based on relative humidity and temperature, and then the depression of the dew point could be calculated based on the temperature and the dew point temperature. The globe data is from 1954 to 2007, with a spatial resolution of 2.5°× 2.5°.
In order to analyze the effects of ENSO and AO, we defined the El Niño events (Tables 1 and 2, positive ESNO), the +AO events (Tables 3 and 4, positive AO), the La Niña events ( Table 2, negative ESNO), and the −AO events (Tables 3, 4 and 5, negative AO) according to the maximum and minimum values of the multivariate ENSO Index (MEI.v1) and a winter AO Index (NOAA Climate Prediction Center). Finally, fourteen El Niño events, fourteen La Niña events, fourteen +AO events, and fourteen −AO events were obtained. In this paper, the composite analysis method was used to explore the climate background during the years with more winter fog days, as well as that during the years with less winter fog days. Based on the observation data of fog days and the re-analysis data, the effects of ENSO and AO on the wintertime fog days in east China were discussed.
Moreover, the influences of ENSO and AO on the meteorological conditions (circulations, temperature, water vapor, etc.) were also analyzed in this paper. Eastern China refers to (110-120°E, 20-40°N); North China refers to (110-120°E, 30-40°N); South China refers to (110-120°E, 20-30°N), and the Yangtze River basin is around 30°N.  Figure 1 shows that from November to February, the number of fog days in eastern China is over 10; the number of fog days in the Yangtze River basin can reach more than 15, and in parts of Southwest and South China, the number can reach more than 20. From Fig. 1b, it can be seen that wintertime fog days account for more than 40% of the annual fog days in eastern China. While in North China, Southwest China, the middle reaches of the Yangtze River basin, and some parts of South China, the ratio can even reach 50% or more. Therefore, it is necessary to analyze the fog days in eastern China in wintertime where and when the fog occurs frequently.
Previous studies (Yu et al. 2019;Liu et al. 2020a) have shown that ENSO and AO are important influence factors on fog days. But, are there any differences between ENSO and AO? Besides, previous studies also showed that the ENSO impacts in early winter and late winter are quite different, so it is necessary to divide the wintertime into early winter (November-December; abbreviated as ND) and late winter (January-February; abbreviated as JF). These differences between early and late winter may be related to the seasonal evolution (Liu et al. 2020b). However, are the effects of AO on fog days different between in ND and in JF? To answer this question, Fig. 2 discusses the performance of the two indexes in ND and JF.
To remove the effect of inter-decadal variations and trends, the 9-point smoothing result is subtracted from the original index, and thus, the inter-annual signal is reserved. Figure 2a and Fig. 2b show the ENSO and AO indexes  in ND and JF, respectively. Their auto-correlation coefficients can reach 0.96 and 0.30, respectively. The ENSO index has very strong inter-annual variation, and the auto-correlation between the ENSO indexes in ND and JF is also large. But, the AO index is an atmospheric signal with weak persistence, so the auto-correlation of the AO index between ND and JF is weak. However, there is a significant difference in the strength of the inter-annual oscillation of the AO index in ND before and after 1980. It is weak before 1980 and strong after 1980. While the AO index in JF presents significant inter-annual variation over the whole period. Figure 2c shows the auto-correlation coefficients of the two indexes in adjacent months. It can be found that the autocorrelation coefficients of the ENSO index are all above 0.9, with the largest in Oct-Feb, especially in ND when it reaches the peak of 0.99. But, the persistence of high auto-correlation is weak, and the auto-correlation decreases after winter. So, it is also an obstacle for ENSO forecasting. Since the ENSO signal has a better persistence in winter, the ENSO index in ND is used as an influencing factor on fog days in eastern China. It can be seen in Fig. 1c that the monthly persistence of the AO index auto-correlation is weak, and it only becomes a little persistent in winter and spring. Because of the weak persistence, it is necessary to use AO indexes in ND and JF respectively to study its impact on fog days. The correlation coefficient of the AO index between ND and JF is 0.3, which means that on a bimonthly average, it is and would be persistent. It can be seen from Table 1 that most of the correlation coefficients have not passed the significance test. AO is positively correlated to ENSO in ND but negatively correlated to ENSO in JF, indicating that the relationship between AO and ENSO is changed in ND and JF. Especially in JF, the AO is significantly related to ENSO, possibly due to the fact that ENSO has an effect on the atmospheric signal AO after reaching its peak in the winter and receives the response from AO. Therefore, based on the above relationship between ESNO and AO and the results of previous studies (Yu et al. 2019;Liu et al. 2020a), the combined effects of ENSO and AO on fog days in eastern China in winter will be studied.
In this section, based on the ENSO index, fourteen El Niño years and fourteen La Niña years are selected. Figure 3a and Fig. 3b respectively show the difference of fog days between El Niño years and La Niña years in ND and JF in eastern China. It can be seen that the difference in the fog days between ND and JF is huge in El Niño or La Niña years. In ND, fog days are more frequent in eastern China, centered in North China, while they are less frequent in JF. Figure 3c and Fig. 3d are the difference of fog days between AO positive anomaly (+AO) years and AO negative anomaly (−AO) years in eastern China in ND and JF, respectively. In +AO years, the distribution of fog days in ND and JF are basically the same, with more fog days in North China and less fog days in South China. In order to discuss the effects of ENSO and AO more fairly, the 14-year composite data is used for each analysis. But, the influence of AO on fog days is more significant in JF than in ND. The comparison reveals that ENSO has a greater impact on fog days in ND (Fig. 3a), while AO has a greater impact in JF (Fig. 3d).  Figure 4 shows the standard deviation distribution of winter fog days in eastern China in ND andJF during 1954-2007. It is shown that the whole eastern China has fog days of more than 2-3 days, with the number in ND larger than in JF. North China, Southwest China, and South China are the regions with a large variation of the fog days, which can reach over 4 days in ND and over 3 days in JF.
Further comparison between the two factors shows that, in ND, the effects of ENSO in North China can account for over 40% of the standard deviation of fog days, and in JF, its effects on most parts of eastern China are basically less than 20%. In ND, AO in North China can explain more than 20-30% of the standard deviation of fog days, but it is weaker in South China. In JF, the AO impact is significantly stronger and can reach over 40% in North China and 20-30% in South China.
In the following, the impacts of ENSO and AO on fog days will be analyzed through the atmospheric circulations.

The relationships between the fog days in eastern China and the atmospheric circulations in the years of ENSO or AO
The latitude vertical profiles of the water vapor and wind fields (Fig. 5) show that water vapor is more affected by ENSO and less affected by AO. From the climatology of vertical velocity, North China is in the region of sinking airflow, and the atmospheric stratification is stable there. So, the key to the occurrence of fog days is whether the water vapor supersaturation could occur. Under the influence of El Niño, 25°N is the center of the anomalous increase in water vapor during ND. In 30-40°N, the water vapor increase and the sinking airflow are conducive to the increase of fog days in North China. While in JF, the positive anomaly of moisture weakens in 25°N, so the number of fog days does not increase Water vapor supersaturation is a necessary condition for the occurrence of fog. As seen from the above figures, in the case of the positive anomaly of ENSO and AO, the warming and wetting occur simultaneously. Therefore, to achieve the water vapor supersaturation, the increasing rate of water vapor must be faster than that of temperature, and we could not just consider the variation of water vapor or temperature.
From Fig. 6, it can be seen that the areas with negative dew point deficit (T−Td) are easy to supersaturate. When El Niño occurs, the anomalous warm-wet airflow transports a large amount of water vapor northward, and the transport is large in ND and slightly less in JF. Due to the faster increase in water vapor and the slower increase in temperature in ND, a significant negative dew point deficit appears in 20-50°N in eastern China (Fig. 6a), leading to a significant increase of fog days (Fig. 3a). While in JF, the transport of anomalous warm-wet airflows to the north is slightly weaker. Although there is a significant increase in temperature, the increase in water vapor is slower, and negative dew point deficit only occurs in 20-30°N. Accompanied by a weak downdraft, the number of fog days increases slightly in that region. In 35°N of North China, the dew point deficit reaches the positive maximum value of 0.6 K (Fig. 6b), and the number of fog days decreases due to the drier atmosphere.
When a positive AO anomaly occurs, there is a significant temperature increase in North China due to the weakening of the East Asian trough, while the water vapor increase is insignificant. During ND, although the increase in water vapor is weak, there are some areas where the increase of water vapor is faster than that of temperature. In 32-42°N, with negative dew point deficit and sinking airflows, more fog days are likely to occur. While in 20-30°N, the vertical air motion is dominated by updrafts and the dew point deficit is positive, so the fog days are fewer. In JF, due to the northward movement of the warming center, most parts of eastern China are in a negative temperature anomaly. But, the water vapor is increased in 20-40°N, and then, a large-value area of negative dew point  Figure 7a and Fig. 7b show the difference of composite 850 hPa wind and specific humidity anomaly between El Niño and La Niña years in ND and JF. As can be seen, the ENSO event occurs with an anomalous warm-wet southwesterly airflow, which is larger in ND and slightly smaller in JF. In ND, there is a significant anomalous high in Japan at the 500-hPa height field, accompanied by stronger anomalous 850-hPa southwesterly airflows, leading to significant warming and wetting in eastern China. However, in ND, the wetting is stronger (Fig. 7a) and the warming is weaker (Fig.  8a), so there is a significant increase in the precipitation in eastern China where the center of anomalous precipitation coincides with the center of the anomalous water vapor increase (Fig. 9a). Then, the anomalous sinking airflow would occur on the north of the 27°N precipitation center, causing more fog days in North China. In JF, however, the anomalous southwesterly airflow is weaker (Fig. 7b), so is the ability of the warm-wet airflow to advance northward. At a 500-hPa height field, there is an anomalous high pressure in Siberia, resulting in significantly higher warming in eastern China than in ND (Fig. 8b). Thus, the large-value center of the Fig. 7 The difference of composite 850 hPa wind and specific humidity anomaly (shaded) between El Niño and La Niña years in a ND and b JF (vector, unit: m/s; shading, unit: g/kg). c and d are the same as a and b except for between +AO and −AO years precipitation anomaly in Fig. 9b appears more southerly (Fig.  9b). Around 20°N in North China, the increasing rate of warming is higher than wetting, so the supersaturation is less likely to occur, resulting in less fog days.
When a positive AO anomaly occurs, the East Asian trough would weaken, and a significant increase in the height field and a temperature increase would appear in North China, but the water vapor increase is not significant. During ND, on the north of 30°N (Fig. 7c), there is anomalous easterly wind that are actually the weakened climatic low-level jet, resulting in a water vapor increase in North China. Anomalous high pressure occurs at 50°N in eastern China at 500 hPa, and there is an anomalous warm center in the low levels (Fig. 6c), with significant warming in eastern China (Fig. 8c). Although there are few areas with significant water vapor increase, some areas in 32-42°N where the water vapor increases faster than temperature and the downdraft prevails are prone to have more fog days.
As can be seen in Fig. 6c, there is a negative dew point deficit at 20°N and 35°N, indicating that both regions are prone to supersaturation, which corresponds well to the two precipitation centers (Fig. 9c). The descending motion in 35°N of North China is still significant, so the fog days there increase. In JF, the weakening of the East Asian trough is more pronounced, the anomalous easterly wind is stronger, and the low-level jet is more intense. As a result, the water vapor anomaly in North China is more significant (Fig. 7d). Although the warming is stronger (Fig. 8d), the warming center has moved northward, and most of eastern China is of a negative temperature anomaly. At 30°N, the increased moisture and supersaturation cause more precipitation in the Yangtze River basin (Fig. 9d). The precipitation increase leads to the temperature decrease in the lower atmosphere. The supersaturation in North China interacts with the downdrafts, and as a result, the number of fog days increases significantly in North China. Nevertheless, South China has a significant (c) (d) (a) (b) Fig. 8 The difference of composite 500 hPa height and T anomaly (shaded) between El Niño and La Niña years in a ND and b JF (contour, unit: gpm; shading, unit: K). c and d are the same as a and b except for between +AO and −AO years increase in precipitation and strong updrafts, which results in a decrease of fog days in South China.

The combined effects of ENSO and AO on wintertime fog days in eastern China
To further illustrate the combined effects of ENSO and AO on fog days in eastern China, we select the AO positive years out of the fourteen El Niño years and fourteen La Niña years (Table 5). Then, the composite analysis is given in Fig. 10 and Fig. 11. The fog days during El Niño years (Fig. 10) are Fig. 9 The difference of composite rainfall anomaly between El Niño and La Niña years in a ND and b JF (solid, unit: mm/month). c and d are the same as in a and b except for between +AO and −AO years. Shading represents the values exceeding the 90% confidence level Nov-Dec Jan-Feb +AO El Nino 1972, 1982, 1994, 20061959, 1964, 1973+AO La Nina 1975, 1988, 19991957, 1976, 1989, 2000 Composite of anomalous fog days over eastern China in a ND and b JF during El Nino years, respectively. c and d are the same as a and b except for during +AO years. e and f are the same as a and b except for both +AO and El Nino years. Shading represents the values exceeding the 90% confidence level Fig. 11 Composite of anomalous fog days over eastern China in a ND and b JF during La Nina years, respectively. c and d are the same as a and b except for during +AO years. e and f are the same as a and b except for both +AO and La Nina years. Shading represents the values exceeding the 90% confidence level almost half of the difference of composite fog days between El Niño and La Niña years (Fig. 3). In ND, the fog days affected only by El Niño are compared with the fog days affected by El Niño and positive AO anomaly (Figs. 10a, 10c, and 10e). It can be seen that there are many fog days in North China under the two situations. However, affected by the positive AO anomaly, insignificantly less fog days appear in South China. Then, the fog days affected by La Niña alone are compared with the fog days affected by La Niña and positive AO anomaly ( Fig. 11a and 11e). The distributions of fog day anomaly are similar in the two cases, with fog days less frequent in eastern China. The above analyses show that in ND, the influence of ENSO is dominant, and the influence of AO on fog days is limited. The combined effects of ENSO and AO are far greater than the sum of their individual effects (Figs. 10a,10c,10e,11a,11c,and 11e). In JF, the fog days affected only by El Niño and the fog days affected by both El Niño and positive AO anomaly are compared (Fig. 10b, 10d, and 10f). It can be seen that in the two cases, North China has more fog days and South China has less fog days. The distribution of fog days affected by both El Niño and positive AO anomaly is very similar to that affected by only the positive AO anomaly. Then, the distributions of fog days affected by La Niña alone, by the positive AO anomaly alone, and affected by both La Niña and positive AO anomaly are compared (Fig. 11b, 11d, and 11f). It can be seen in Fig. 11f that, in eastern China, there are more fog days in North China and fewer fog days in South China, but this phenomenon is not significant. The above analyses show that in JF, AO is basically dominant, and the effect of ENSO on fog days is limited. Especially in JF, the combined effects of ENSO and AO are very obvious. Whether in El Niño or La Niña years, the combined effects of ENSO and AO are far greater than individual AO effects in Northern China (Figs. 10d,10f,11d,and 11f).
In ND in the years of both El Niño and positive AO anomaly (Fig. 12a), the Japanese high at 500 hPa is strengthened, and there is a warming in eastern China. The southwesterly warm-moist airflow could bring abundant water vapor to eastern China (Fig. 13a). However, the supersaturation and a negative dew point deficit appear in South China (Fig. 14a). With the strong updrafts, heavy precipitation events frequently occur in South China. However, the supersaturation near 40°N is accompanied by the sinking airflow, which increases the fog days in North China. The effect of the positive AO anomaly is the warming in eastern China. When the water vapor is constant, fewer areas are prone to supersaturation under the warming. Thus the area with frequent fog days in eastern China is reduced. When El Niño and positive AO anomaly work together in JF, the Siberian high anomaly enhances due to the strong influence of positive AO anomaly (Fig. 12b). Although in El Niño years there is a weakening of the southwesterly airflow during JF, the southwesterly flow could enhance anomalously due to the positive AO anomaly (Fig. 15b). Consequently, the water vapor increase is very significant, and the wetting is more rapid than warming in North China. As a result, supersaturation is more likely to occur in North China (Fig. 14b). Accompanied by the sinking airflow, the fog days of North China increase significantly. Besides, the strong southwesterly moisture transport (Fig. 13b) increases the heavy precipitation in South China and reduces the possibility of fog day occurrence. In ND in the years of both La Niña and positive AO anomaly, although the high-pressure anomaly and temperature increase occur in the areas north of 40°N (Fig. 12c), the temperature in most parts of eastern China is lower and the southwesterly warm-wet airflow changes to the negative anomaly. With the cold-dry air from the north moving southward, the water vapor is in a negative anomaly (Fig. 13c), and it is indicated that the atmosphere is getting colder and drier (Fig.  15c). Moreover, the dew point deficit in most parts of eastern China is in a positive anomaly (Fig. 14c), and the air is unlikely to supersaturate. Therefore, the fog days in eastern China are reduced. In contrast, the positive AO anomaly in ND has basically no influence on the areas south of 40°N. In JF of the La Niña years, the anomalous height centered on Fig. 13 The same as Fig. 5, but a and b are for both +AO and El Nino years in ND and JF, respectively. c and d are for both +AO and La Nina years in ND and JF Japan at 500 hPa increases, and the East Asian trough weakens significantly (Fig. 12d). Under the influence of the strong AO anomaly in JF, the anomalous southeasterly wind in eastern China strengthens and brings an anomalous increase of water vapor (Fig. 15d). Besides, La Niña is conducive to the cooling in eastern China. The decrease in temperature and the increase in water vapor causes a negative anomaly in dew point deficit over a wide area of eastern China (Fig. 14c). Accompanied with the sinking airflow in North China, a large-value center of fog days then appears at 35°N. Yet with the strong ascending movement (Fig. 13d), South China becomes the large-value center of precipitation, without significant change of fog days. Therefore, in La Niña years, the significant difference in fog days between ND and JF is mainly caused by the intensity of AO.

Discussion and conclusions
In this paper, the seasonal characteristics and spatial distribution of wintertime fog days in eastern China are analyzed, and the individual and combined effects of ENSO and AO on the meteorological background fields (circulation field, temperature, water vapor, etc.) related to the fog days in eastern China. The conclusions are as follows.
First, the fog days in eastern China are more frequent in winter, accounting for more than 40-50% of the whole year, so it is important to study the fog days in winter. As an important tropical climatic impact factor, ENSO plays significantly different roles during ND and JF in eastern China, affecting the extent and intensity of fog days. Besides, the impacts of AO are also distinctly different in ND and JF. Therefore, wintertime here is divided into ND and JF.
During the ND of the ENSO-positive phase, fog days are frequent in eastern China, and the main large-value area is in North China. During the JF of the ENSO-positive phase, ENSO has little influence on the fog days in eastern China, and the number of fog days in North China even decreases slightly. During the ND of the ENSO-positive phase, there is a stronger southwesterly warm-wet airflow, which increases the temperature and humidity in eastern China, and the increasing rate of wetting is faster than warming. Thus, the whole eastern China is prone to reach water vapor supersaturation. In South China, there are strong upward motions and precipitation Fig. 14 The same as Fig.6, but a and b are for both +AO and El Nino years in ND and JF, respectively. c and d are for both +AO and La Nina years in ND and JF, respectively positive anomalies. However, in North China, there are more fog days due to the climatic sinking airflow and the water vapor supersaturation. In JF, the southwesterly warm-wet airflow is weaker. Although there is precipitation in South China, the intensity is weaker than that in ND. Besides, the warm-wet airflow is difficult to move northward to North China, so there are slightly less fog days.
AO has significant effects on fog days in eastern China, both in ND and JF. In the positive phase of AO, the number of fog days increases in North China and decreases in South China, and this phenomenon is more pronounced in JF than in ND. Meanwhile, with the polar pressure weakening, the middle-and low-level pressure in East Asia anomalously enhances. The East Asian trough weakens, and the westerly low-level jet moves northward, preventing northwesterly cold air from moving southward. Warming and wetting in eastern China lead to limited regions prone to water vapor supersaturation, which are mainly concentrated in North China. Thus, fog days increase in North China. The increase in water vapor is limited, and the overall precipitation anomaly in eastern China is smaller than the precipitation anomaly induced by ENSO. The water vapor in South China decreases, and fog days become fewer in South China. In contrast, the AO positive anomaly is more pronounced in ND than in JF. With the weaker Fig. 15 The same as Fig.7, but a and b are for both +AO and El Nino years in ND and JF, respectively. c and d are for both +AO and La Nina years in ND and JF